Storage Overview

Authkestra is entirely Database Agnostic. We never enforce database schemas, migrations, or ORM choices.

The Trait-Based Approach

All data persistence is defined through simple Rust traits. For example, to manage sessions, the engine relies on the SessionStore trait:

#[async_trait]
pub trait SessionStore: Send + Sync {
    async fn get(&self, id: &str) -> Option<Session>;
    async fn set(&self, session: Session) -> Result<(), AuthError>;
    async fn delete(&self, id: &str) -> Result<(), AuthError>;
}

Because this is a trait, you can implement it using sqlx, diesel, redis, or a simple in-memory HashMap.

Included Stores

Authkestra comes with a few generic implementations out of the box for convenience. To use them, you must enable the corresponding feature flags in your Cargo.toml:

[dependencies]
# Example: Using Redis and SQLite stores
authkestra = { version = "0.2.1", features = ["redis", "sql-sqlite"] }

The available built-in stores and their feature flags are:

• Memory Store: authkestra_engine::store::memory::MemoryStore (Great for testing and local dev; enabled by the memory feature).
• Redis Store: authkestra_engine::store::redis::RedisStore (Production-ready distributed cache; enabled by the redis feature).
• SQL Store: authkestra_engine::store::sql::SqlStore (Relational persistence; enabled by sql-postgres, sql-mysql, or sql-sqlite features).